amba-pl011/dma: Add check for the residue in DMA callback
In DMA-operated uart, I found that rx data can be taken by the UART
interrupts during the DMA irq handler. pl011_int is occurred just
before it goes inside spin_lock_irq. When it returns to the callback,
DMA buffer already has been flushed. Then, pl011_dma_rx_chars gets
invalid data. So I add check for the residue as the patch bellow.
Signed-off-by: Chanho Min <chanho.min@lge.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
diff --git a/drivers/tty/serial/amba-pl011.c b/drivers/tty/serial/amba-pl011.c
index 6800f5f..cc3ea06 100644
--- a/drivers/tty/serial/amba-pl011.c
+++ b/drivers/tty/serial/amba-pl011.c
@@ -827,7 +827,12 @@
{
struct uart_amba_port *uap = data;
struct pl011_dmarx_data *dmarx = &uap->dmarx;
+ struct dma_chan *rxchan = dmarx->chan;
bool lastbuf = dmarx->use_buf_b;
+ struct pl011_sgbuf *sgbuf = dmarx->use_buf_b ?
+ &dmarx->sgbuf_b : &dmarx->sgbuf_a;
+ size_t pending;
+ struct dma_tx_state state;
int ret;
/*
@@ -838,11 +843,21 @@
* we immediately trigger the next DMA job.
*/
spin_lock_irq(&uap->port.lock);
+ /*
+ * Rx data can be taken by the UART interrupts during
+ * the DMA irq handler. So we check the residue here.
+ */
+ rxchan->device->device_tx_status(rxchan, dmarx->cookie, &state);
+ pending = sgbuf->sg.length - state.residue;
+ BUG_ON(pending > PL011_DMA_BUFFER_SIZE);
+ /* Then we terminate the transfer - we now know our residue */
+ dmaengine_terminate_all(rxchan);
+
uap->dmarx.running = false;
dmarx->use_buf_b = !lastbuf;
ret = pl011_dma_rx_trigger_dma(uap);
- pl011_dma_rx_chars(uap, PL011_DMA_BUFFER_SIZE, lastbuf, false);
+ pl011_dma_rx_chars(uap, pending, lastbuf, false);
spin_unlock_irq(&uap->port.lock);
/*
* Do this check after we picked the DMA chars so we don't